Regional Clusters: More Complex Than You Think

Economists have noted for over a century that similar firms tend to “cluster” near each other. In the last few decades, research on clusters has picked up dramatically, in part because they are associated with more rapid innovation. One of the best-known examples of a contemporary regional cluster is Silicon Valley and its cluster of electronics and software firms.

Clusters are often assumed to work due to openness, collaboration, loose organizational boundaries, and information sharing. A new research paper* by Simon Bell, Paul Tracey, and Jan Hiede argues that it’s often more complex than this. They call this “standard” model of a cluster “a relational model of governance based on implicit rules and understandings.” And they point out that there is a lot of evidence for another type of successful cluster: one that’s hierarchically organized, with “unilateral rules originating from a dominating firm” (they cite several examples of research showing this, from the U.S. to England to Shanghai). The problem is that most research has only focused on the relational type of cluster, within which “innovation is an interactive and collective process requiring joint action on the part of cluster members” with “organic” interaction patterns between cluster members. A hierarchical cluster, in contrast, has “centralized decision making structures, rules, and formalized interactions.”

So their research question is, how does a cluster evolve into one or the other form? What variables make one more likely than another? The authors argue for two causal variables:

(1) information tacitness: tacit information is not codified and is difficult to communicate explicitly. It often involves unspoken practices and behaviors. The authors propose that the greater extent of tacit information in a cluster, the more likely it is to be relational.

(2) transaction specific investments: Someone has to pay to build the factories, machinery, hire the workers, and build a brand. In some industries (think automobiles) this costs a lot more than in others (think creating a web site). The authors propose that the greater the degree of specific investments required, the more likely the cluster is to be hierarchical.

The authors identify four other propositions that are too complex to summarize here, but this is an intriguing paper. In many ways, although the authors don’t say it this way, it challenges the mystique of clusters as being nice, good, and collaborative. (After Saxenian’s book comparing Silicon Valley to Boston’s Route 128 came out, it undeniably made Silicon Valley look friendlier and happier and made Route 128 look old-fashioned and uptight.) For example, on p. 630 the authors note:

For many of the key players in the global electronics industry, the motivation to belong to geographical clusters stems from a desire to control information flows between plants and to reduce spillovers, rather than from a desire to create new knowledge through a process of interactive learning.

This is a long a somewhat complex academic article, but it will reward close reading.

*Bell, Tracey, and Heide. 2009. The organization of regional clusters. Academy of Management Review, Vol. 34 No. 4, 623-642.